BEN RAPHAEL
courtesy Ben Raphael

His finalist year: 1992

His finalist project: Teaching a computer to simulate how insects might learn to walk

What led to the project: Ben Raphael was "very into artificial intelligence" and robots in general as a child. But rather than becoming obsessed with Isaac Asimov novels as many kids do, "I was more science nerd than a science fiction nerd," he says. He read as much as he could about neural networks—basically, complex computer programs that simulate how the brain and nervous systems solve problems. He learned to program the computer his parents bought for their Fairfax, Va., home in the early 1980s and, as a student at Paul VI Catholic High School, decided to do a science project that showed what these machines could do.

His plan? To build a computer neural network that would simulate how a population of insects would develop the ability to walk. "My idea—kind of naively as a high school student—was, if we want to teach robots to walk, how does it really happen in nature?" he says. He studied the literature to learn how two- and four-legged animals strut and gallop, then looked at six-legged creatures.

When he plugged equations describing how animals walk into his computer, and gave the insects with the ability to walk the farthest the ability to survive to reproduce, over generations the insects' legs became more controlled and synchronized. The software allowed users to watch the process on the screen. Given the state of graphics more than 15 years ago, "I'm sure it would look cooler now," he says, but the program was good enough to earn him a finalist spot in the 1992 Westinghouse Science Talent Search.

The effect on his career: Many Westinghouse and Intel finalists end up doing work completely different from their projects. Not Raphael, who wound up building computer models of biological processes as a career.